On April 14 Beijing Time, Demis Hassabis, CEO of DeepMind, made startling judgments about the current state and future direction of artificial intelligence in a recent podcast interview. He stated outright that AGI (Artificial General Intelligence) is expected to be achieved within five years, while warning that current AI is in a special phase where it is "overhyped in the short term and seriously underestimated in the long term."

The True Scale of the AI Revolution: A Tenfold Industrial Revolution Happening Ten Times Faster Hassabis emphasized that this AI wave is not an ordinary technological iteration but a "tenfold industrial revolution happening ten times faster." The side effects of the industrial revolution took humanity a century to absorb, but this time, all impacts will be compressed into just ten years. He quantified AGI as a "tenfold industrial revolution at ten times the speed," reminding the industry to prepare in advance for large-scale social and economic shocks.

The Gap Among the Top Four AI Companies Is Widening, Not Shrinking Contrary to the common belief that the gap is narrowing, Hassabis pointed out that the technological gap between the world's top four AI companies is continuing to widen. This means the future competitive landscape will become more fragmented, with the advantages of the leading companies possibly becoming even more entrenched.

Current AI Is 'Patchy Intelligence' Hassabis described the core feature of current AI as "patchy intelligence" — when asked in a certain way, AI performs surprisingly well; but when phrased differently, the same question can fail at the most basic level. He believes the fundamental reason is that AI is still fragmented and cannot weave all knowledge into a coherent whole like the human brain. This flaw leads to obvious shortcomings in AI's reliability in complex scenarios.

Scaling Laws Still Work, But the Rules of the Game Are Changing Regarding the widely discussed Scaling Laws, Hassabis said that the phenomenon of diminishing returns does exist, but "it's not dead yet." The real change lies in the fact that the competition has shifted from the past "who has more money and computing power" to "who can come up with new ideas." Innovation will become the key factor determining success in the next stage.

Editorial Comment from AIbase